@genai-fi/nanogpt 0.19.0 → 0.20.0

package/dist/ones-BytntneX.js

@@ -1,14 +0,0 @@
-import { u as n, V as t, U as m, E as i } from "./index-CUXkjxiT.js";
-import { c as f } from "./complex-DI35Q-gW.js";
-import { z as c } from "./zeros-DvZpK8s6.js";
-function l(o, r = "float32") {
-  if (n(o), r === "complex64") {
-    const s = l(o, "float32"), a = c(o, "float32");
-    return f(s, a);
-  }
-  const e = t(m(o), r);
-  return i.makeTensor(e, o, r);
-}
-export {
-  l as o
-};

package/dist/ops/adamAdjust.d.ts

@@ -1,2 +0,0 @@
-import { Tensor } from '@tensorflow/tfjs-core';
-export declare function adamAdjust(moments: Tensor, value: Tensor, beta1: number, beta2: number, epsilon: number, learningRate: number, weightDecay?: number): Tensor;

package/dist/ops/adamAdjust.js

@@ -1,9 +0,0 @@
-import { h as u } from "../index-CUXkjxiT.js";
-import "./cpu/adamAdjust.js";
-import "./webgl/adamAdjust.js";
-function s(r, t, n, e, m, o, i = 0) {
-  return u().runKernel("AdamAdjust", { moments: r, value: t }, { beta1: n, beta2: e, epsilon: m, learningRate: o, weightDecay: i });
-}
-export {
-  s as adamAdjust
-};

package/dist/ops/adamMoments.d.ts

@@ -1,2 +0,0 @@
-import { Scalar, Tensor } from '@tensorflow/tfjs-core';
-export declare function adamMoments(moments: Tensor, gradient: Tensor, beta1: number, beta2: number, scaling: Scalar): Tensor;

package/dist/ops/adamMoments.js

@@ -1,9 +0,0 @@
-import { h as t } from "../index-CUXkjxiT.js";
-import "./cpu/adamMoments.js";
-import "./webgl/adamMoments.js";
-function s(n, r, e, m, o) {
-  return t().runKernel("AdamMoments", { moments: n, gradient: r, scaling: o }, { beta1: e, beta2: m });
-}
-export {
-  s as adamMoments
-};

package/dist/ops/add16.d.ts

@@ -1,2 +0,0 @@
-import { Tensor } from '@tensorflow/tfjs-core';
-export declare function add16(a: Tensor, b: Tensor): Tensor;

package/dist/ops/add16.js

@@ -1,9 +0,0 @@
-import { j as t, h as o } from "../index-CUXkjxiT.js";
-import { isPackedTensor as n } from "../utilities/packed.js";
-import "./grads/add16.js";
-function m(r, e) {
-  return !n(r) && !n(e) ? t(r, e) : o().runKernel("Add16", { a: r, b: e });
-}
-export {
-  m as add16
-};

package/dist/ops/appendCache.d.ts

@@ -1,2 +0,0 @@
-import { Tensor } from '@tensorflow/tfjs-core';
-export declare function appendCache(item: Tensor, maxSize: number, pastLen: number, cache?: Tensor): Tensor;

package/dist/ops/appendCache.js

@@ -1,22 +0,0 @@
-import { h as e } from "../index-CUXkjxiT.js";
-import "./cpu/appendCache.js";
-import "./webgl/appendCache.js";
-import { isPackedTensor as c } from "../utilities/packed.js";
-import { c as t } from "../concat-CAQpCret.js";
-import { z as f } from "../zeros-DvZpK8s6.js";
-function C(r, o, n, p) {
-  if (!p) {
-    const s = r.shape[2], a = c(r);
-    return t(
-      [
-        r,
-        f([r.shape[0], r.shape[1], o - s, r.shape[3]], a ? "int32" : r.dtype)
-      ],
-      2
-    );
-  }
-  return e().runKernel("AppendCache", { cache: p, item: r }, { maxSize: o, pastLen: n });
-}
-export {
-  C as appendCache
-};

package/dist/ops/attentionMask.d.ts

@@ -1,2 +0,0 @@
-import { Tensor } from '@tensorflow/tfjs-core';
-export declare function attentionMask(q: Tensor, k: Tensor, divisor: number, pastLen?: number): Tensor;

package/dist/ops/attentionMask.js

@@ -1,10 +0,0 @@
-import { h as e } from "../index-CUXkjxiT.js";
-import "./cpu/attentionMask.js";
-import "./webgl/attentionMask.js";
-import "./grads/attentionMask.js";
-function u(t, n, o, r) {
-  return e().runKernel("AttentionMask", { q: t, k: n }, { divisor: o, pastLen: r || 0 });
-}
-export {
-  u as attentionMask
-};

package/dist/ops/concat16.d.ts

@@ -1,2 +0,0 @@
-import { Rank, Tensor } from '@tensorflow/tfjs-core';
-export declare function concat16<R extends Rank = Rank>(x: Tensor<R>[], axis?: number): Tensor<R>;

package/dist/ops/concat16.js

@@ -1,9 +0,0 @@
-import { isPackedTensor as o } from "../utilities/packed.js";
-import { h as c } from "../index-CUXkjxiT.js";
-import { c as e } from "../concat-CAQpCret.js";
-function p(r, n) {
-  return o(r[0]) ? c().runKernel("Concat16", r, { axis: n ?? -1 }) : e(r, n);
-}
-export {
-  p as concat16
-};

package/dist/ops/cpu/adamAdjust.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/adamAdjust.js

@@ -1,18 +0,0 @@
-import { c as k, i as t, j as i, m as z, l as A } from "../../index-CUXkjxiT.js";
-function C(c) {
-  const { moments: s, value: r } = c.inputs, { beta1: l, beta2: m, epsilon: u, learningRate: d } = c.attrs, e = s.shape.length, a = new Array(e).fill(0), n = s.shape.slice();
-  n[e - 1] = 1;
-  const o = a.slice();
-  o[e - 1] = 1;
-  const b = n.slice(), p = s.slice(a, n).squeeze([e - 1]), M = s.slice(o, b).squeeze([e - 1]), g = t(p, l), f = t(M, m);
-  return i(
-    z(t(g, i(A(f), u ?? 1e-8)), -d),
-    r
-  );
-}
-const h = {
-  kernelName: "AdamAdjust",
-  backendName: "cpu",
-  kernelFunc: C
-};
-k(h);

package/dist/ops/cpu/adamMoments.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/adamMoments.js

@@ -1,16 +0,0 @@
-import { c as p } from "../../index-CUXkjxiT.js";
-import { s as b } from "../../stack-DudVrtmG.js";
-function f(t) {
-  const { moments: n, gradient: c } = t.inputs, { beta1: o, beta2: m } = t.attrs, e = n.shape.length, a = new Array(e).fill(0), s = n.shape.slice();
-  s[e - 1] = 1;
-  const i = a.slice();
-  i[e - 1] = 1;
-  const r = s.slice(), l = n.slice(a, s).squeeze([e - 1]), u = n.slice(i, r).squeeze([e - 1]), M = l.mul(o).add(c.mul(1 - o)), d = u.mul(m).add(c.square().mul(1 - m));
-  return b([M, d], -1);
-}
-const g = {
-  kernelName: "AdamMoments",
-  backendName: "cpu",
-  kernelFunc: f
-};
-p(g);

package/dist/ops/cpu/appendCache.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/appendCache.js

@@ -1,23 +0,0 @@
-import { c as d } from "../../index-CUXkjxiT.js";
-import { c as h } from "../../concat-CAQpCret.js";
-function u(p) {
-  const { cache: n, item: s } = p.inputs, { maxSize: i, pastLen: c } = p.attrs, t = n.shape[0], o = n.shape[1], a = n.shape[3], e = s.shape[2];
-  if (c + e <= i) {
-    const f = n.slice([0, 0, 0, 0], [t, o, c, a]), m = n.slice([0, 0, c + e, 0], [t, o, i - c - e, a]), r = e < e ? s.slice([0, 0, 0, 0], [t, o, e, a]) : s, k = h([f, r, m], 2);
-    return f.dispose(), m.dispose(), r !== s && r.dispose(), k;
-  }
-  const l = n.slice([0, 0, e, 0], [t, o, i - e, a]), C = h([l, s], 2);
-  return l.dispose(), C;
-}
-const w = {
-  kernelName: "AppendCache",
-  backendName: "cpu",
-  kernelFunc: u
-};
-d(w);
-const N = {
-  kernelName: "AppendCache",
-  backendName: "tensorflow",
-  kernelFunc: u
-};
-d(N);

package/dist/ops/cpu/attentionMask.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/attentionMask.js

@@ -1,22 +0,0 @@
-import { c as o, f as d, a as u } from "../../index-CUXkjxiT.js";
-import { l as N } from "../../ops-CsXeTq1P.js";
-import { o as b } from "../../ones-BytntneX.js";
-import { z as A } from "../../zeros-DvZpK8s6.js";
-import { w as I } from "../../not_equal-hurPF26l.js";
-import { m as g } from "../../mat_mul-DhG0Newp.js";
-function a(n) {
-  const { q: s, k: e } = n.inputs, { divisor: r } = n.attrs, c = s.shape[2], t = e.shape[2], m = N.bandPart(b([t, t]), -1, 0).cast("bool"), i = A([t, t]), l = d([t, t], Number.NEGATIVE_INFINITY), f = I(m, i, l), k = g(s, e, !1, !0).mul(u(r)), p = f.slice([0, 0], [c, t]).expandDims(0).expandDims(0);
-  return k.add(p);
-}
-const w = {
-  kernelName: "AttentionMask",
-  backendName: "cpu",
-  kernelFunc: a
-};
-o(w);
-const M = {
-  kernelName: "AttentionMask",
-  backendName: "tensorflow",
-  kernelFunc: a
-};
-o(M);

package/dist/ops/cpu/fusedSoftmax.d.ts

@@ -1,9 +0,0 @@
-import { SoftmaxAttrs, SoftmaxInputs, TensorInfo } from '@tensorflow/tfjs-core';
-interface FusedSoftmaxAttrs extends SoftmaxAttrs {
-    dropoutRate?: number;
-}
-export declare function softmaxCPU(args: {
-    inputs: SoftmaxInputs;
-    attrs: FusedSoftmaxAttrs;
-}): TensorInfo;
-export {};

package/dist/ops/cpu/fusedSoftmax.js

@@ -1,29 +0,0 @@
-import { c as e } from "../../index-CUXkjxiT.js";
-import { s as m } from "../../softmax-D4q1LJN7.js";
-function n(t) {
-  const { inputs: s, attrs: a } = t, { logits: o } = s, { dim: i, dropoutRate: r } = a;
-  if (!o)
-    throw new Error("Error in softmax: input logits is null");
-  return r !== void 0 && r > 0 && console.warn("Dropout in fusedSoftmax not implemented for CPU backend, skipping dropout."), m(o, i);
-}
-const f = {
-  kernelName: "FusedSoftmax",
-  backendName: "cpu",
-  kernelFunc: n
-};
-e(f);
-const u = {
-  kernelName: "FusedSoftmax",
-  backendName: "tensorflow",
-  kernelFunc: n
-};
-e(u);
-const c = {
-  kernelName: "FusedSoftmax",
-  backendName: "webgpu",
-  kernelFunc: n
-};
-e(c);
-export {
-  n as softmaxCPU
-};

package/dist/ops/cpu/gatherSub.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/gatherSub.js

@@ -1,18 +0,0 @@
-import { o as u, q as c, E as g, al as h, c as l, a1 as m } from "../../index-CUXkjxiT.js";
-import { r as p } from "../../range-C-CjF-LI.js";
-import { s as N } from "../../stack-DudVrtmG.js";
-function f(e, t) {
-  const n = c(t, "indices", "gatherND", "int32"), s = { params: c(e, "x", "gatherND", "string_or_numeric"), indices: n };
-  return g.runKernel(h, s);
-}
-const b = /* @__PURE__ */ u({ gatherND_: f });
-function d(e) {
-  const { values: t, labels: n, logits: r } = e.inputs, s = n.shape[0], a = p(0, s, 1, "int32"), o = N([a, n], 1), i = b(r, o);
-  return m(t, i);
-}
-const k = {
-  kernelName: "EfficientGatherSub",
-  backendName: "cpu",
-  kernelFunc: d
-};
-l(k);

package/dist/ops/cpu/gelu.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/gelu.js

@@ -1,40 +0,0 @@
-import { c as t, t as d } from "../../index-CUXkjxiT.js";
-const o = 0.7978845608028654, c = 0.044715;
-function m(r) {
-  const { inputs: u } = r, { x: n } = u, e = n;
-  return d(() => {
-    const l = e.pow(3), s = e.add(l.mul(c)).mul(o).tanh().add(1).mul(0.5);
-    return e.mul(s);
-  });
-}
-const p = {
-  kernelName: "Gelu",
-  backendName: "cpu",
-  kernelFunc: m
-};
-t(p);
-const K = {
-  kernelName: "Gelu",
-  backendName: "tensorflow",
-  kernelFunc: m
-};
-t(K);
-function i(r) {
-  const { dy: u, x: n } = r.inputs;
-  return d(() => {
-    const e = n.square(), l = e.mul(n), a = n.add(l.mul(c)).mul(o).tanh(), f = a.square().neg().add(1), k = e.mul(3 * c).add(1), g = a.add(1).mul(0.5), G = n.mul(f).mul(o).mul(k).mul(0.5), N = g.add(G);
-    return u.mul(N);
-  });
-}
-const x = {
-  kernelName: "GeluGrad",
-  backendName: "cpu",
-  kernelFunc: i
-};
-t(x);
-const h = {
-  kernelName: "GeluGrad",
-  backendName: "tensorflow",
-  kernelFunc: i
-};
-t(h);

package/dist/ops/cpu/matMul16.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/matMul16.js

@@ -1,15 +0,0 @@
-import { isPackedTensor as t } from "../../utilities/packed.js";
-import { c as p } from "../../index-CUXkjxiT.js";
-import { m as l } from "../../mat_mul-DhG0Newp.js";
-function m(e) {
-  const { A: n, B: r } = e.inputs, { transposeA: o, transposeB: s } = e.attrs, a = !t(n), c = !t(r);
-  if (a && c)
-    return l(n, r, o, s);
-  throw new Error("MatMul16 CPU kernel only supports packed tensors currently.");
-}
-const u = {
-  kernelName: "MatMul16",
-  backendName: "cpu",
-  kernelFunc: m
-};
-p(u);

package/dist/ops/cpu/matMulGelu.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/matMulGelu.js

@@ -1,53 +0,0 @@
-import { c as e, t as m } from "../../index-CUXkjxiT.js";
-import { g as M, d as i } from "../../gelu-B220X1Go.js";
-import { m as k } from "../../mat_mul-DhG0Newp.js";
-function c(t) {
-  const { inputs: u } = t, { x: n, kernel: r } = u, a = n, l = r;
-  return m(() => {
-    const o = k(a, l);
-    return M(o);
-  });
-}
-const G = {
-  kernelName: "MatMulGelu",
-  backendName: "cpu",
-  kernelFunc: c
-};
-e(G);
-const f = {
-  kernelName: "MatMulGelu",
-  backendName: "tensorflow",
-  kernelFunc: c
-};
-e(f);
-const p = {
-  kernelName: "MatMulGelu",
-  backendName: "webgpu",
-  kernelFunc: c
-};
-e(p);
-function s(t) {
-  const { dy: u, x: n, kernel: r } = t.inputs;
-  return m(() => {
-    const a = k(n, r), l = i(u, a), o = l.matMul(r.transpose()), d = n.transpose().matMul(l);
-    return [o, d];
-  });
-}
-const g = {
-  kernelName: "MatMulGeluGrad",
-  backendName: "cpu",
-  kernelFunc: s
-};
-e(g);
-const N = {
-  kernelName: "MatMulGeluGrad",
-  backendName: "tensorflow",
-  kernelFunc: s
-};
-e(N);
-const b = {
-  kernelName: "MatMulGeluGrad",
-  backendName: "webgpu",
-  kernelFunc: s
-};
-e(b);

package/dist/ops/cpu/matMulMul.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/matMulMul.js

@@ -1,23 +0,0 @@
-import { c as e, t as i } from "../../index-CUXkjxiT.js";
-function n(t) {
-  const { inputs: r, attrs: l } = t, { transposeA: o, transposeB: s } = l, { x: u, kernel: c, y: a } = r, m = u, k = c, M = a;
-  return i(() => m.matMul(k, o, s).mul(M));
-}
-const p = {
-  kernelName: "MatMulMul",
-  backendName: "cpu",
-  kernelFunc: n
-};
-e(p);
-const f = {
-  kernelName: "MatMulMul",
-  backendName: "tensorflow",
-  kernelFunc: n
-};
-e(f);
-const g = {
-  kernelName: "MatMulMul",
-  backendName: "webgpu",
-  kernelFunc: n
-};
-e(g);

package/dist/ops/cpu/mulDropout.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/mulDropout.js

@@ -1,23 +0,0 @@
-import { c as e, m as t } from "../../index-CUXkjxiT.js";
-function n(o) {
-  const { inputs: r } = o, { a: l, b: u } = r;
-  return console.warn("Using fallback mulDrop implementation without dropout."), t(l, u);
-}
-const c = {
-  kernelName: "MulDropout",
-  backendName: "cpu",
-  kernelFunc: n
-};
-e(c);
-const a = {
-  kernelName: "MulDropout",
-  backendName: "tensorflow",
-  kernelFunc: n
-};
-e(a);
-const m = {
-  kernelName: "MulDropout",
-  backendName: "webgpu",
-  kernelFunc: n
-};
-e(m);

package/dist/ops/cpu/normRMS.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/normRMS.js

@@ -1,39 +0,0 @@
-import { c as o, t as d } from "../../index-CUXkjxiT.js";
-function i(t) {
-  const { inputs: e } = t, { x: n, gamma: s } = e, r = n, a = s;
-  return d(() => {
-    const u = r.square().mean(-1, !0).add(1e-8).rsqrt();
-    return r.mul(u).mul(a);
-  });
-}
-const f = {
-  kernelName: "RMSNorm",
-  backendName: "cpu",
-  kernelFunc: i
-};
-o(f);
-const g = {
-  kernelName: "RMSNorm",
-  backendName: "tensorflow",
-  kernelFunc: i
-};
-o(g);
-function N(t) {
-  const { dy: e, x: n, gamma: s } = t.inputs;
-  return d(() => {
-    const r = n.shape[n.shape.length - 1], a = n.square().mean(-1, !0), m = a.add(1e-8).rsqrt(), u = n.mul(m), l = e.mul(u).sum([0, 1]), c = e.mul(s), k = c.mul(n).sum(-1, !0).div(r);
-    return [c.mul(m).sub(n.mul(k).mul(m).div(a.add(1e-8))), l];
-  });
-}
-const S = {
-  kernelName: "RMSNormGrad",
-  backendName: "cpu",
-  kernelFunc: N
-};
-o(S);
-const R = {
-  kernelName: "RMSNormGrad",
-  backendName: "tensorflow",
-  kernelFunc: N
-};
-o(R);

package/dist/ops/cpu/qkv.d.ts

@@ -1,5 +0,0 @@
-import { NamedAttrMap, NamedTensorInfoMap, TensorInfo } from '@tensorflow/tfjs-core';
-export declare function qkvCPU(args: {
-    inputs: NamedTensorInfoMap;
-    attrs?: NamedAttrMap;
-}): TensorInfo[];

package/dist/ops/cpu/qkv.js

@@ -1,41 +0,0 @@
-import { c as q } from "../../index-CUXkjxiT.js";
-import { r as o } from "../../reshape-BDOuCSNW.js";
-import { s as x } from "../../split-C2Sj255c.js";
-function v(p) {
-  const { x: c, kernel: K } = p.inputs, { heads: n, packed: C } = p.attrs;
-  if (C)
-    throw new Error("QKV CPU implementation does not support packed tensors.");
-  const [s, e, t] = c.shape, a = o(c, [s * e, t]), i = a.dot(K);
-  a.dispose();
-  const d = o(i, [s, e, 3 * t]);
-  i.dispose();
-  const [k, l, m] = x(d, 3, -1);
-  d.dispose();
-  const r = t / n, f = o(k, [s, e, n, r]);
-  k.dispose();
-  const w = f.transpose([0, 2, 1, 3]);
-  f.dispose();
-  const h = o(l, [s, e, n, r]);
-  l.dispose();
-  const N = h.transpose([0, 2, 1, 3]);
-  h.dispose();
-  const u = o(m, [s, e, n, r]);
-  m.dispose();
-  const T = u.transpose([0, 2, 1, 3]);
-  return u.dispose(), [w, N, T];
-}
-const F = {
-  kernelName: "QKV",
-  backendName: "cpu",
-  kernelFunc: v
-};
-q(F);
-const Q = {
-  kernelName: "QKV",
-  backendName: "tensorflow",
-  kernelFunc: v
-};
-q(Q);
-export {
-  v as qkvCPU
-};

package/dist/ops/cpu/rope.d.ts

@@ -1,6 +0,0 @@
-import { NamedAttrMap, NamedTensorInfoMap, Tensor, TensorInfo } from '@tensorflow/tfjs-core';
-export declare function applyRoPE(sinCache: Tensor, cosCache: Tensor, rotaryDim: number, q: Tensor, pastLen: number): Tensor;
-export declare function ropeCPU(args: {
-    inputs: NamedTensorInfoMap;
-    attrs?: NamedAttrMap;
-}): TensorInfo;

package/dist/ops/cpu/rope.js

@@ -1,38 +0,0 @@
-import { c as I } from "../../index-CUXkjxiT.js";
-import { r as y } from "../../range-C-CjF-LI.js";
-import { g as F } from "../../gather-qIqEqaGn.js";
-import { s as E } from "../../stack-DudVrtmG.js";
-import { c as T } from "../../concat-CAQpCret.js";
-function U(c, r, p, e, n) {
-  const t = e.shape[3], s = p;
-  if (s > t) return e;
-  const o = e.shape[2], i = s / 2, a = r.slice([n, 0, 0], [o, i, 1]).reshape([1, 1, o, i]), d = c.slice([n, 0, 0], [o, i, 1]).reshape([1, 1, o, i]), l = e.shape[0], m = e.shape[1], h = y(0, s, 2, "int32"), g = y(1, s, 2, "int32"), D = ((k) => {
-    const C = k.slice([0, 0, 0, 0], [l, m, o, s]), R = s < t ? k.slice([0, 0, 0, s], [l, m, o, t - s]) : null, u = F(C, h, 3), f = F(C, g, 3), v = u.mul(a), N = f.mul(d), S = v.sub(N), P = f.mul(a), b = u.mul(d), x = P.add(b);
-    u.dispose(), f.dispose(), a.dispose(), d.dispose(), v.dispose(), N.dispose(), P.dispose(), b.dispose();
-    const K = E([S, x], -1);
-    S.dispose(), x.dispose();
-    const w = K.reshape([l, m, o, s]);
-    return K.dispose(), R ? T([w, R], 3) : w;
-  })(e);
-  return h.dispose(), g.dispose(), D;
-}
-function B(c) {
-  const { x: r } = c.inputs, { pastLen: p, negSin: e, ropeCache: n } = c.attrs, t = r.shape[3], s = e ? n.getNegSin() : n.getSin(), o = n.getCos();
-  return U(s, o, t, r, p);
-}
-const j = {
-  kernelName: "Rope",
-  backendName: "cpu",
-  kernelFunc: B
-};
-I(j);
-const z = {
-  kernelName: "Rope",
-  backendName: "tensorflow",
-  kernelFunc: B
-};
-I(z);
-export {
-  U as applyRoPE,
-  B as ropeCPU
-};

package/dist/ops/cpu/scatterSub.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/ops/cpu/scatterSub.js

@@ -1,23 +0,0 @@
-import { o as f, u as g, q as r, E as l, ak as N, c as b, a1 as S, m as h } from "../../index-CUXkjxiT.js";
-import { v as k } from "../../scatter_nd_util-ByNJaL6I.js";
-import { r as D } from "../../range-C-CjF-LI.js";
-import { s as v } from "../../stack-DudVrtmG.js";
-import { o as E } from "../../ones-BytntneX.js";
-function I(a, e, s) {
-  g(s);
-  const n = r(a, "indices", "scatterND", "int32"), t = r(e, "updates", "scatterND");
-  k(t, n, s);
-  const c = { indices: n, updates: t }, o = { shape: s };
-  return l.runKernel(N, c, o);
-}
-const K = /* @__PURE__ */ f({ scatterND_: I });
-function L(a) {
-  const { logits: e, labels: s, dy: n } = a.inputs, t = s.shape[0], c = e.shape[1], o = D(0, t, 1, "int32"), i = v([o, s], 1), d = E([t]), u = K(i, d, [t, c]), p = S(e, u), m = n.reshape([t, 1]);
-  return h(p, m);
-}
-const T = {
-  kernelName: "EfficientScatterSub",
-  backendName: "cpu",
-  kernelFunc: L
-};
-b(T);

package/dist/ops/dot16.d.ts

@@ -1,2 +0,0 @@
-import { Tensor } from '@tensorflow/tfjs-core';
-export declare function dot16(a: Tensor, b: Tensor, transposeA?: boolean, transposeB?: boolean): Tensor;